Skip to main content

Chemical News

Para-Chloro Was Good Enough For Them, So It’s Good Enough For Me

How many of the molecular pieces that we use in medicinal chemistry are historical accidents? I’ve wondered this from time to time. There’s no doubt that drug structures are partly driven by ease of synthesis/commercial availability (these two go hand in hand), and these in turn are influenced by which reactions and feedstocks were exploited earlier. The Grignard reaction came well before palladium coupling methods, but there’s no reason that it had to, just to pick one example.
This new paper in J. Med. Chem. is what has me thinking about this again. The authors, from AstraZeneca, show that their in-house chemists tend to think of para-aromatic substituents more often than the other regioisomers, and that this preference is mirrored in the commercially available reagents (and indeed, in marketed drugs). The paper looks into sources of this bias – cost, the 1972 Topliss tree paper, and so on, but no single factor appears to be at work. What does seem to be going on is a self-reinforcing bias – there are more p-aromatics in the screening deck, so more of them hit. And there are more commercially available compounds with the structure, so more of them get made in turn.

We believe that ultimately the present day bias is now likely due to unjustified personal preferences and overused at the expense of meta and ortho regioisomers as well as other potentially diverse bioisosteres. This last point is an important conclusion. The bias for p-ClPh has propagated throughout the years and influenced design and synthesis plans. A simple extension into disubstituted aromatics revealed that chemists favor the similarly substituted compounds (e.g., diCl, diF, diMeO), with many of these having at least one element in the para position. This analysis also illustrated that many disubstituted compounds are underrepresented in the public domain, highlighting an opportunity for screening collection differentiation.

I suspect that there are many more such biases, based on availability of different heterocycles, lack of stereoselective methods in some areas, etc. Our screening collections (and our building block catalogs) are the work of human beings, making conscious and unconscious choices, not some random slice of chemical space.

18 comments on “Para-Chloro Was Good Enough For Them, So It’s Good Enough For Me”

  1. Industry Guy says:

    Isnt their prevalence more likely due to relative inertnness, smallish size and usefulness at blocking para-oxidation??? I know in my years, i have yet to see and monosubstituted phenyl not have one hydroxy metabolite. Weird how these are usually blocked by a para halide. I wonder if in 30 years, will p-chloro prevalence be replaced by p-F?

  2. Industry Guy 2 says:

    Working with an ortho-chloro compound lately, it should be noted that in some cases it is significantly more challenging to make it and its analogs (in a synthetic sense), than it would be to use the corresponding p-chloro compound.

  3. anon the II says:

    I was actually a fan of meta. I figured that it had more ways to bind (2x).
    I’ve always felt that when someone (usually from AZ in the UK) does these types of analysis, they short-change the role of synthetic expediency. Synthetic expediency explains more things than a terabyte of QSAR.
    Oh yeah, people don’t do ortho much cause it don’t work much.

  4. Anonymous says:

    @3. anon the II
    You said, “…people don’t do ortho much cause it don’t work much.”
    The tables show that meta is used just as infrequently, even though it doesn’t have the same synthetic accessibility problems that ortho might have; therefore, the reason for the their infrequent use must be something else.

  5. A Nonny Mouse says:

    One of the chemists at BW at RTP always declared meta chloro to be the one; we always thought that he was biased as his name was “Mehta”(Bupropion inventor).

  6. eugene says:

    It’s racism really. No matter how they try to spin the word ‘bias’. The metas have been proven by science to be inferior, so I’ve got no argument with that, but the orthos are just as good as the paras, even though their steric effects get in the way at times and make you angry. Articles like this may mean well, but in order to keep drug discovery functioning we need to restrict the migration of these meta (and even ortho) substituents into our libraries. Metas are deactivators and they and other ‘diverse bioisosteres’ will ruin drug discovery, you mark my words!
    Why do you think there are only paradontists and orthodontists anyways?

  7. Pete says:

    My ‘regio-prejudice’ is that para substitution with chloro will give better protection against metabolism than either meta or ortho substitution although will concede that I have never put this to the test by performing the appropriate matched molecular pair analysis. When putting fragment screening libraries together, I favor meta substitution because this gives pairs of conformers of essentially the same energy (buy one molecular shape, get one free) although some ligand detect NMR folk prefer para for spectroscopic reasons. Ortho substitution tends to have conformational effects although, if memory serves me correctly, you usually have to substitute both ortho positions. I favor chloro as a substituent in fragment libraries because it seems to be a bit ‘stickier’ than methyl.

  8. Bill says:

    And if our selection of diversity libraries and synthetic design continue to be influenced by computational models built from our existing collections data, the bias will remain and continue to expand.

  9. will says:

    that is a question for metaphysicians

  10. Al Gore says:

    If I was an AZ shareholder I might be wondering why exactly it was that these scientists had nothing better to do with their time than write this paper.

  11. milkshake says:

    Para isomers are often the main product of chlorinations, and they crystallize better so it is easier to get them pure… (This reminds me: p-tosyl- became the cheapest and most readily available sulfonyl group very early on due to saccharine manufacturing (the required o-tosylchloide for making saccharine is actually the minor product of toluene chlorosulfonylation.)

  12. Farmhand says:

    It is interesting that the graphic shows the highest para-bias in known drugs, even higher than the reagent and screening libraries. I am just curious if this is again the bias of medicinal chemists or something else? Are the authors suggesting that all the way to the clinic chemists are just not making the other substitution patterns and blindly following what’s in the library? I started to write down the structures that I have first-hand knowledge of that have been identified as strong leads and a few that were identified for at least some follow-up beyond discovery. The sample size of compounds in my “personal” library that go beyond that is too small to be meaningful. For each of these more advanced compounds, I then tried to recall how we got there. In this very small sample, p-halogen only slightly outnumber m-halogen. O-halogen is absent, but o-N and o-C are well represented, albeit not as much as p-halogen. In some cases, the halogen was added after initial screening to address a real or perceived PK or Tox issue. In others, it was in the screening library but was swapped out or moved as part of optimization. Its presence or absence in the screening library had no impact. The ortho substitutions were a mix of screening and optimization additions. In most cases, o-substitution had a profound impact on potency, either greatly increasing or decreasing, so it was a bit risky beyond the screening stage, but when successful was considered brilliant (rarely my idea, I must admit). Bottom line from my personal vantage point: I am sure there is a chemistry-based para-bias in screening libraries that impacts how diverse a sample you really get, but I am not so sure that this has had any meaningful impact on the ultimate identification of leads or development compounds. Some of this bias (ortho-substitution) may be founded in passed bad experience. Maybe we should use FEP (later post) to see if there is any systematic bias in nature, since that’s the one that counts.

  13. Harry says:

    Based only on Derek’s post, I cannot follow the logic of this paper. Any bias in a screening library likely reflects commercially available compounds (as most companies fill their screening sets largely from purchase). If optimization plans do not probe alternative aromatic substitutions of halogens beyond para that might turn up as hits, what are the medicinal chemists at AZ doing? The quote from the paper in the post does not reflect kindly on discovery chemists at AZ.
    What matters is not the distribution of aryl chlorides in the screening deck, but rather what has been made as part of optimization campaigns.

  14. DrSnowboard says:

    @10 as an AZ shareholder you’d be happy that the site is in the process of being flogged off or shut down, with a historical bias towards the latter. Unfortunately, it does highlight the disconnection between departments, because , like, they could have asked a chemist why.

  15. Handles says:

    The main reason I use chloro-substituted building blocks is because the mass spec is so distinctive. Count how many chlorines are in each peak in the LCMS and you can often get a quick understanding of how the chemistry works, or doesnt work.

  16. skatesailor says:

    Bias against (or for) ortho- and meta-substituents may arise because they can impart undesirable (or desirable) chirality, depending on the size, shape, and proximity of the R-group in R-C6H4-Cl. Even meta-substituted aromatic rings can create atropisomers isolable at room temperature (Chirality 8, 364-371 (1996)).

  17. Yet another Scientist says:

    What I find interesting is that meta is underrepresented in drugs compared to all the rest….

  18. A Nonny Mouse says:

    Too true- we had this problem with some lamotrigine analogues (ortho, meta dichloro).

Comments are closed.